This object is used to provide an additional level of priority that is required for certain queuing disciplines and when the different queues that comprise a queue set are serviced using a mix of queuing disciplines. This object can be used to specify, for example, the order in which queues will be serviced when priority queuing is used. It also supports the ability to describe the servicing hierarchy when a hybrid queuing scheme, such as priority queuing coupled with weighted fair queuing, is used.

Queue service priority is assigned such that a lower service order value indicates a higher priority. For example, a priority queue with a value of 1 will be serviced (i.e., drained) before another priority queue with a service order value of 2.

Note that multiple queues that are logically associated, based on the queuing discipline that is being employed, will be assigned the same service order value. Under this scenario, other parameters that are related to the queuing discipline determine the order of queue servicing (e.g., queue drain size is used for 'wfq').

For example, an interface that is associated with a queue set supporting two priority queues and three queues that are serviced using WFQ would be modeled as follows:

Q Index Q Discipline Q Drain Size Q Service Order 22 pq(1) - 1 23 pq(1) - 2 24 wfq(3) 500 3 25 wfq(3) 350 3 26 wfq(3) 150 3

The queue set presented in this example would service all queued traffic in queue 22 first, followed by all of the queued traffic in queue 23. Next the queued traffic in queues 24 through 26 would be serviced in a round robin fashion with queue 24 receiving 50% of the available bandwidth, queue 25 receiving 35% of the available bandwidth and queue 26 receiving 15% of the available bandwidth. This example is presented for expository purposes and has been simplified accordingly.

Note that, in this example, queues 24, 25 and 26 form a queue cluster. Members of a queue cluster are all assigned the same qosIfQueueServiceOrder as there are tightly coupled. The qosIfQueueDrainSize attribute is used to determine the additional processing characteristics of the individual queues in a cluster.